Association of Age With Treatment-Related Adverse Events and Survival in Patients With Metastatic Colorectal Cancer

Key Points Question Is there any difference in treatment-related adverse events and outcomes between patients with early-onset (age <50 years) and older patients with metastatic colorectal cancer (mCRC)? Findings In this cohort study of 1223 patients with mCRC from 3 clinical trials, early-onset mCRC had significantly worse survival and displayed unique patterns of treatment-related adverse events. Analysis in a separate Moffitt Cancer Center cohort of 736 patients noted worse overall survival in early-onset mCRC and showed distinct genomic profiles in this population, which may partially explain the disparity. Meaning These findings might have utility for an individualized approach to chemotherapy, counseling, and management of treatment-related adverse events in patients with early-onset mCRC.


Introduction
Colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancerrelated death worldwide. 1 Despite the decrease of overall incidence and mortality of CRC as a result of nationwide-adopted screening programs, 2-4 the incidence of the early-onset CRC (EO-CRC) in patients diagnosed with CRC before age 50 years has been increasing by 2% annually since the 1990s. 5,6 It is projected that by 2030 10.9% of all colon cancers and 22.9% of all rectal cancers will be diagnosed as EO-CRC. 7,8 Neither the etiology of increased incidence nor the distinct biology of EO-CRC compared with its older counterparts is clearly revealed. For example, inherited cancer syndromes, such as Lynch syndrome, may be more prevalent in patients with EO-CRC. However, only 5% to 7% of those with EO-CRC carry deleterious germline mutations of mismatch repair genes. 9,10 Previous studies examining the genomic profile of EO-CRC showed few molecular differences between EO-CRC and its older counterparts. 11 Even the few molecular differences reported were often not confirmed in other studies likely due to selection bias in different patient cohorts. 8,12,13 Some studies investigated nongenetic risk factors for CRC, such as low physical activity, excess alcohol consumption, smoking, and obesity, but the results were neither conclusive nor adequate explanations of the increased incidence of EO-CRC. [14][15][16] The increased incidence of EO-CRC poses unique challenges to the management of treatment in younger patients, especially with regard to the prognosis and treatment-related adverse events. 17 In current clinical practice, patients with EO-CRC receive similar, if not more aggressive, therapies in the advanced disease setting compared with their older counterparts. Fluoropyrimidine-based combination chemotherapy plus biologic agents remains the most common first-line therapy for patients with metastatic colorectal cancer (mCRC). 18 Triplet chemotherapy combination is an alternative for patients who are younger and have more aggressive disease and good performance status, offering a better outcome at the expense of worse toxic effects. 19 Although chemotherapy in combination with biologic agents 20,21 showed comparable benefit in patients with advanced CRC of all ages, 22,23 it is unclear whether there is disparity in treatment-related adverse events and outcome from advanced CRC between EO and its older counterparts. Previous studies suggested an increased incidence of some specific toxic effects, such as nausea and vomiting in EO-mCRC, but the data were largely limited in the adjuvant setting and findings across studies were not always in agreement. [24][25][26][27][28] Similarly, some studies reported a poorer survival in patients with advanced EO-CRC, 24,25,29 whereas others observed a similar survival outcome between patients with EO-mCRC and those with average onset. 26-28,30,31 These discrepancies might be due to selection bias and/or the inclusion of older patients (age >65 years) in the average-onset group for comparison.
To address this literature gap, we used individual patient data from 3 clinical trials in which patients with mCRC received first-line fluorouracil and oxaliplatin (FOLFOX) therapy and a contemporary patient cohort to compare the differences of treatment outcome and adverse events among 3 different age groups of patients with mCRC.

Methods
We used individual patient data from 3 multicenter randomized phase 3 clinical trials in the Project Data Sphere, which were divided into study 1 (NCT00272051, NCT 00305188) and study 2 (NCT00364013). Study 1 assessed the efficacy of xaliproden in reducing (NCT00272051) or preventing (NCT 00305188) the neurotoxic effects of FOLFOX as first-line treatment for patients with mCRC. Study 2 evaluated the efficacy of panitumumab combined with FOLFOX as first-line therapy for patients with mCRC. Only patient data from the control (FOLFOX only) arms of the 3 clinical trials were combined and used in our study. Study design, inclusion and exclusion criteria, interventions, end points, adverse events, and outcomes of the 3 clinical trials have been previously

Statistical Analysis
Our statistical analysis was performed using R Statistical Software, version 4.1.2 (R Foundation for Statistical Computing) along with the survival package 3.2-13. The χ 2 test and Fisher exact test were used to test the distribution of baseline characteristics, adverse events, and NGS genomic features in the 3 groups. Continuous variables were compared with the Kruskal-Wallis test and χ 2 or Fisher exact test were performed for categorical variables. Overall survival and PFS were evaluated according to the Kaplan-Meier method and compared using the log-rank test. A Cox proportional hazards regression model was used to determine the association of adverse events with OS and PFS. P values in this study were 2-sided, and P < .05 was considered statistically significant. Pairwise comparisons adjusting for multiple testing were performed using the Benjamini and Hochberg method.

Demographic and Clinical Characteristics
Among all 1959 patients (1145 [58.4%] men; 814 [41.6%] women) included, 1223 were from study 1 and study 2 for analyses of survival outcomes and treatment-related adverse events. A total of 736 patients were from the Moffitt cohort for assessment of OS and tumor genetic alterations. As presented in the Table, the Moffitt cohort included a significantly higher rate of patients with EO-mCRC (26.6%) compared with study 1 (15.2%) and study 2 (13.7%) (P < .001), but a lower prevalence of White patients (83.3%) vs those of other races and ethnicities (16.7%) (P < .001). Of the patients from study 1 and study 2, 179 (14.6%) were younger than 50 years, 582 (47.6%) were aged 50 to 65 years, and 462 (37.8%) were older than 65 years at the time of stage IV CRC diagnosis (eTable 1 in Supplement 1). There were significantly more women younger than 50 years in contrast to more men older than 65 years. Fewer White patients and more individuals of other races and ethnicities were represented in the age less than 50 years group. No significant difference of Eastern Cooperative Oncology Group (ECOG) performance status was observed among the 3 age groups (eTable 1 in Supplement 1).

Age-Related Disparity of Treatment-Related Adverse Event Pattern
Examination of treatment-related adverse events of all patients from study 1 and study 2 combined revealed a unique pattern for patients younger than 50 years ( Figure 2). As shown in Figure 2A, B and  Figure 2C and eTable 4 in Supplement 1. In addition, patients younger than 50 years had a shorter duration of mucositis (0.6 vs 0.9 vs 1.0 weeks; P = .006) as shown in Figure 2D and eTable 5 in Supplement 1.

Association of Treatment-Related Adverse Events With OS and PFS
As summarized in Figure 3A and   were associated with poor PFS. In contrast, severe peripheral neuropathy (HR, 0.43; 95% CI, 0.26-0.71; P = .001) was associated with better PFS ( Figure 3B; eTable 6 in Supplement 1).
Patients younger than 50 years were more likely to receive triplet therapy and carry a left-sided tumor, while those older than 65 years had a higher rate of patients who were White. Despite the difference of baseline characteristics between the MCC cohort and study 1 and 2 (Table), patients younger than 50 years consistently showed worse median OS compared with those aged 50 to 65 years and comparable median OS to those older than 65 years (39.2 vs 51.3 vs 38.0 months; P = .02) ( Figure 4A). Further examination of genetic alterations was performed to explore the potential underlying factors in this age-related disparity of survival outcome in patients with mCRC. As presented in Figure 4B and eTable 8 in Supplement 1, we found that all 3 age groups had very similar genetic alterations except that, compared with the other 2 age groups, the tumors of patients

Discussion
A hypothesis for an increased incidence of EO-CRC despite a decrease in the overall incidence 5,6 was that EO-CRC may be a different disease as reflected by distinct clinicopathologic and molecular features. 5,39 Previous studies that attempted to investigate such age-related disparity of survival outcomes had discordant findings. Several studies found an association between younger age and worse survival, 25,29,40,41 whereas other studies revealed opposite findings. 24 For example, in the study of Lieu and colleagues, 29 when age was treated as a continuous variable, older age (age >65 years) was associated with worse survival, presumably due to reduced life expectancy, increased comorbidities, or inability to tolerate aggressive chemotherapy. 47 Therefore, in our study, we divided patients into 3 age groups (<50, 50-65, and >65 years). As expected, we found that patients both younger than 50 and those older than 65 years had worse OS compared with those aged 50 to 65 years. Another distinct feature of our study is that patients received the same treatment, first-line FOLFOX, to minimize possible risk modification from different treatment  To use excellent recording of treatment-related adverse events in clinical trial data, we also compared these data among the 3 age groups and observed age-related disparity in patients with mCRC. Consistent with previous findings in the adjuvant setting, 24-28 our data showed an increased incidence of nausea and vomiting but a decreased incidence of diarrhea, fatigue, and neutropenia in younger patients. The previously reported potential improvement of the toxic effect profile in younger patients 27,50 was not entirely supported by our study, as we observed a higher incidence of severe abdominal pain, severe anemia, and severe rash in patients with EO-mCRC. These are in line with the findings from the IDEA trials, which revealed an increased incidence of severe nausea and vomiting in younger patients. 25 In addition, EO-mCRC had an earlier onset of nausea and vomiting, mucositis, and neutropenia. Hence, it is more appropriate to say that patients with EO-mCRC exhibited a unique pattern of adverse events after receiving first-line FOLFOX therapy. Moreover, the association between severe abdominal pain and severe liver toxic effects with worse survival in younger patients suggested an individualized approach to the monitoring and management of these unique treatment-related adverse events.
The age-related disparity of survival and adverse events may suggest unique underlying disease biologic factors in different age groups. Comprehensive genomic profiling in some studies noted a higher prevalence of TP53 and CTNNB1 mutations, but a lower prevalence of BRAF and APC mutations in EO-mCRC, 11,28,43,51,52 which is consistent with some findings in our study.
The CTNNB1 and CREBBP genes are involved in the Wnt/β-catenin pathway, 53 and ERBB2 amplification is involved in the MAPK/ERK pathway. 54 Thus, our study suggests that these pathways might be implicated in worse survival outcomes in EO-mCRC. However, it is still unclear whether these differential genetic alterations might partially explain the observed disparity in survival outcome given the inconsistent findings across studies and the numerically small differences in few of many genes. Therefore, further efforts should likely focus on multiomic studies to inform disease biologic factors and therapy.

Limitations
This study has limitations. The 3 clinical trials took place in an era during which biologic agents had not become standard first-line therapy for patients with mCRC and clinical NGS was not used to guide treatment. These clinical trials were conducted with FOLFOX treatment before more intensive therapy, such as folinic acid, fluorouracil, oxaliplatin and irinotecan, became available. Furthermore, they did not collect information on treatment intensity, adherence, and location and number of metastases, which prevented us from adjusting for these factors in our analysis. To compensate and address these limitations intrinsic to the data used in our study, we incorporated institutional patient and clinical NGS data from the modern era, encompassing a more diverse range of patients and treatments, as an external validation cohort. Distributions of some important baseline characteristics, such as race and ethnicity, tumor sidedness, and first-line treatments in this MCC patient cohort were consistent with those previously reported. 26,28,31,48 However, we acknowledge a limitation in not evaluating adverse events, PFS, or conducting multivariable survival analysis in this cohort due to the following concerns: retrospective data collection, even when sourced from prospectively maintained databases, is susceptible to selection bias and inaccuracy; and the highly heterogeneous nature of this patient population concerning multiple genes, treatments, and histopathologic characteristics may lead to overfitting and nonconvergence of the multivariable model, given the relatively small sample size of this patient cohort. Consequently, we used Moffitt data solely for external validation of OS, rather than for the primary analysis. In addition, these findings may not generalize to other racial and ethnic categories given that the majority of trial participants were identified as White individuals.